CN1701287A - Interactive device - Google Patents

Abstract

An interactive apparatus 1 which is able to decide on an action pattern in accordance with health conditions of a user without a necessity of putting a biometric sensor on a human body is provided. The interactive apparatus 1 comprises: detection means 50b for detecting a health condition of a user; deciding means 50c for deciding on an action pattern in accordance with the health condition of the user; execution instructing means 50g for instructing execution of the action pattern; offering means 50e for making an offer of the action pattern to the user with a speech before instructing execution of the action pattern; and determination means 50f for determining whether an answer of the user to the offered action pattern is an answer to accept the offered action pattern or not. The execution instructing means 50g instructs execution of the offered action pattern when the answer of the user is determined to be the answer to accept the offered action pattern.

Description

conversational device

technical field

The present invention relates to a dialog device capable of having a dialog with a user.

Background technique

A disclosed audio device monitors life status information, and gradually adjusts the audio signal preferred by the occupant to a level suitable for the occupant's current life status and health status for playback (for example, refer to Patent Document 1). The audio device uses the sensors installed in the room to grasp the living conditions of the occupants, monitors the identification signals and living status information sent by the portable transceivers (including biosensors) equipped by the occupants, and gradually integrates the occupants' preferences. The audio signal is adjusted to be played at a level suitable for the occupant's current state of life and health.

Patent Document 1: Japanese Unexamined Patent Application Publication No. 11-221196

However, in the prior art described in the above-mentioned Patent Document 1, there is such a problem that in order to obtain biological information, etc., the resident needs to be equipped with a portable transceiver, which is troublesome and inconvenient for the resident, and the resident's Life is constantly monitored by sensors placed in the room, which makes people feel uncomfortable.

An object of the present invention is to provide an interactive device capable of determining an action mode according to a user's health state without equipping a human body with a biosensor.

Contents of the invention

The conversational device of the present invention has: a grasping unit for grasping the health state of the user; a decision unit for determining an appropriate action pattern according to the health state of the user grasped by the grasping unit; an execution instruction unit for to instruct the execution of the action pattern determined by the determination unit; the suggestion unit is used to suggest the above-mentioned action pattern to the user through voice before instructing the execution of the action pattern determined by the above-mentioned determination unit; The reply of the suggested action mode judges whether it has accepted the above-mentioned suggested action mode; mode of action, thereby achieving the above-mentioned purpose.

The grasping unit may also grasp the health status of the user according to the user's language expression.

The grasping unit may also grasp the health status of the user according to the keywords spoken by the user.

It is also possible to further have a suggestion necessity determination unit, which is used to determine whether the above-mentioned action mode needs to be suggested to the above-mentioned user before instructing to execute the action mode determined by the above-mentioned decision unit; When it is necessary to suggest the action pattern to the user, the suggestion unit suggests the action pattern to the user by voice.

The advice necessity determination unit may determine the necessity of the advice based on a flag value indicating the necessity of the advice assigned in advance to the action pattern.

The suggestion necessity determination unit may also determine the necessity of the suggestion according to the time distribution of execution times of the above-mentioned action pattern.

The determination unit may determine one of the plurality of action modes assigned with priorities as an action mode suitable for the health state of the user, and adjust the action mode assigned to the action mode according to whether the user accepts the action mode. order of priority.

It may further have a storage unit for storing an action pattern suitable for the health state of the user; the determination unit uses the action pattern stored in the storage unit to determine the action pattern.

The action pattern suggested to the user by the suggestion unit may also include selection of content played in the playback device.

The content may also include audio data, image data and lighting control data, and the playing device changes at least one of light quantity and light color of the lighting device according to the lighting control data.

The above-mentioned interactive device may have at least one of an agent function and a mobile function.

The health status of the user may represent at least one aspect of the user's emotion and the user's health status.

The interactive device of the present invention has: a voice input unit, which is used to convert the voice emitted by the user into a voice signal; a voice recognition unit, which recognizes the language spoken by the user based on the voice signal output by the voice input unit; and a dialogue database for Pre-register the language that the user may speak, and save the correspondence between the registered language and the user's health status; the grasping unit compares the language recognized by the voice recognition unit with the language registered in the dialogue database, The comparison results determine the health status of the user, thereby grasping the health status of the user; the determination unit determines the relationship with the grasping unit based on the action pattern table that stores the correspondence between the health status of the user and the action mode of the interactive device. The grasped action pattern adapted to the health status of the user; the execution instruction unit is used to instruct the execution of the action pattern determined by the decision unit; the suggestion unit is used to instruct the execution of the action pattern determined by the decision unit Synthesizing a suggestion sentence based on the output result of the above-mentioned grasping unit and the output result of the above-mentioned decision unit, sending the suggestion of the above-mentioned action pattern to the above-mentioned user through voice; the determination unit judges whether the user accepts the above-mentioned suggested action pattern according to the reply The above-mentioned suggested action mode is confirmed; when the judgment result considers that the reply of the above-mentioned user is a reply of accepting the above-mentioned suggested action mode, the above-mentioned execution instructing unit instructs to execute the above-mentioned suggested action mode, thereby achieving the above-mentioned purpose.

It may also further have: an accepting unit, which is used to accept the action mode suggested by the above-mentioned user against the suggested action mode; a unit for judging whether the above-mentioned dialogue device can execute the above-mentioned counter-suggested action mode; and when judging A unit for updating the correspondence relationship between the health status of the user and the action pattern of the interactive device stored in the action pattern table when the interactive device is considered to be capable of executing the action pattern suggested in the opposite direction.

Brief description of the drawings

FIG. 1 is a diagram showing the appearance of a robot 1 as an example of an interactive device of the present invention.

FIG. 2 is a diagram showing an example of the internal structure of the robot 1 .

FIG. 3 is a diagram showing an example of the relationship between keywords uttered by the user and the health status of the user stored in the dialogue database 140 .

FIG. 4 is a diagram showing an example of the relationship between the user's health status and the behavior pattern of the robot 1 stored in the information database 160 .

FIG. 5 is a flowchart showing an example of a processing procedure for the robot 1 to grasp the user's health status and instruct execution of an action pattern suited to the user's health status.

FIG. 6 is a diagram showing an example of a configuration of a playback device 2100 that enables synchronous playback of audio data and/or video data and lighting control data.

FIG. 7 is a diagram showing an example of the internal configuration of the voice recognition unit 40 .

Fig. 8a is a diagram showing an example of the internal structure of the processing unit 50 shown in Fig. 2 .

FIG. 8b is a diagram showing another example of the internal structure of the processing unit 50 shown in FIG. 2 .

FIG. 8c is a diagram showing still another example of the internal structure of the processing unit 50 shown in FIG. 2 .

FIG. 9 is a diagram for explaining an example of a method of creating a suggestion sentence using the suggestion unit 50e.

FIG. 10 is a diagram showing an example of the internal configuration of the advice necessity determination unit 50d.

FIG. 11 is a diagram showing an example of the structure of the action suggestion necessity table 162 .

The best way to practice the invention

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

The "user's health status" referred to in this specification means at least one of the user's emotion and the user's health status. The so-called "user" refers to the owner of the interactive device.

FIG. 1 shows the appearance of a robot 1 as an example of the interactive device of the present invention. The robot 1 is designed to be able to have a dialogue with a user.

The robot 1 shown in Fig. 1 has a camera 10 equivalent to "eyes", a speaker 110 and antenna 62 equivalent to "mouth", a microphone 30 and antenna 62 equivalent to "ears", and a "head" or "arm". The movable part 180.

The robot 1 may be an autonomous walking robot (mobile robot) having a moving part 160 capable of moving itself, or may be a type of robot that cannot move by itself.

Any mechanical means may be used to enable movement of the robot 1 . For example, the structure of the robot 1 can be forward and backward by controlling the rotation of the rollers arranged on the hands and feet, or it can be a mobile robot such as tire type or foot type. The robot 1 can be a humanoid robot imitating biped upright animals such as human beings, or a pet robot imitating quadruped animals.

In addition, although an interactive robot has been described as an example of an interactive device, the interactive device is not limited thereto. An interactive device may be any device having a structure capable of conducting a dialog with a user. The interactive device may be, for example, an interactive toy, or an interactive mobile device (including a mobile phone), or an interactive agent.

It is preferable that a conversational agent can act in an information space such as the Internet, and have functions of information processing such as information retrieval, filtering, and schedule adjustment instead of humans (function of a software agent). Conversational agents talk to people like people, so they are also called anthropomorphic agents.

The interactive device may have at least one of an agent function and a mobile function.

FIG. 2 shows an example of the internal structure of the robot 1 .

The image recognition unit 20 acquires an image from the camera 10 (image input unit), recognizes the acquired image, and outputs the recognition result to the processing unit 50 .

The voice recognition unit 40 acquires voice from the microphone 30 (voice input unit), recognizes the acquired voice, and outputs the recognition result to the processing unit 50 .

FIG. 7 shows an example of the internal structure of the voice recognition unit 40 .

The voice is converted into a voice signal waveform by the voice input unit 30 (microphone). The audio signal waveform is output to the audio recognition unit 40 . The voice recognition unit 40 includes a voice detection unit 71 , a comparison calculation unit 72 , a recognition unit 73 , and a registered voice database 74 .

The voice detection unit 71 extracts the waveform portion of the voice signal waveform input from the voice input unit 30 that satisfies a certain standard as the voice interval that the user actually utters, and outputs the voice signal waveform of this interval to the comparison operation unit 72 as a voice waveform. The certain standard for extracting the voice interval mentioned here may be, for example, that the power of the signal waveform in the frequency range below 1 kHz of the general human voice frequency band is above a certain level.

Voice waveforms of languages that the user can speak are associated with the language and registered in the registered voice database 74 in advance.

The comparison operation unit 72 sequentially compares the sound waveform input from the sound detection unit 71 and the sound waveform registered in the registered sound database 74, calculates the degree of similarity with each sound waveform registered in the registered sound database 74, and outputs the calculation result to Identification unit 73. The method of comparing the two sound waveforms used here may be a method of comparing the total difference of the power components of each frequency after performing frequency analysis such as Fourier transform on the sound waveform, or it may be further carried out in polar coordinates after the frequency analysis. The following cepstrum (cepstrum) parameter or Mel cepstrum (Mel cepstrum) parameter considers the time-scaled DP Matching method. In order to improve the efficiency of the comparison calculation, the voice waveform registered in the registered voice database 74 may be used as a comparison factor (for example, the power component of each frequency) in the comparison calculation unit 72 . In addition, in the registered voice database 74, the voice waveform of the user's unintentional utterance such as coughing or moaning is also registered, and the corresponding language is registered as "unintentional utterance". This makes it possible to distinguish between the user's intentional utterance and unintentional utterance.

The recognition unit 73 detects the voice waveform having the highest similarity among the similarities of the respective voice waveforms input from the comparison operation unit 72, and by determining the language corresponding to the voice waveform detected from the registered voice database 74, the voice is The waveform is converted into characters, and the characters are output to the processing unit 50 . When no large difference is found among the similarities, the input voice is considered to be noise, and conversion from the voice waveform to characters is not performed. Alternatively, it can also be converted to a text such as "noise".

FIG. 8a shows an example of the internal structure of the processing unit 50 shown in FIG. 2 .

The processing unit 50 (processing unit 50a) searches the dialogue database 140 based on the voice recognition result of the voice recognition unit 40, and generates a response sentence. The response sentences are output to the speech synthesis unit 100 . The voice synthesis unit 100 synthesizes the response sentence into voice. The synthesized audio is output from an audio output unit 110 such as a speaker.

The dialog database 140 stores dialog patterns and rules for generating answer sentences. And the relationship between the language (keyword) spoken by the user and the user's health status is also stored in the dialog database 140 .

FIG. 3 shows an example of the relationship between the keywords spoken by the user and the health status of the user stored in the conversation database 140 .

In the example shown in FIG. 3 , the relationship between the keywords spoken by the user and the health status of the user is expressed in a tabular form. For example, row 31 of the table indicates that keywords such as "sleepy", "tired" and "no appetite" correspond to the user's health state (physical condition) such as "fatigue". The 32nd row of the table indicates that keywords such as "Good job!" and "Great!" correspond to the user's health status (feeling) such as "joy".

In addition, the method of expressing the relationship between the keywords spoken by the user and the user's health status is not limited to the method shown in FIG. 3 . The relationship between the keywords spoken by the user and the user's health status can be represented by any representation method.

The processing unit 50 (grasp means 50b) extracts a keyword from the voice recognition result of the voice recognition unit 40, and searches the dialog database 140 using the keyword. As a result, the processing unit 50 (ascertainment means 50b) grasps the health status of the user based on the keyword. For example, when the keyword extracted from the voice recognition result is one of "sleepy", "tired" and "no appetite", the processing unit 50 (grasp unit 50b) refers to the table shown in FIG. "Fatigue" state.

In addition, instead of the above-mentioned method of using keywords, or on the basis of the above-mentioned method of using keywords, the strength or tension of the user's voice can be further grasped according to the voice recognition result, thereby grasping the user's health status. For example, when the processing unit 50 (controlling unit 50b) detects that the intensity or tension of the user’s voice is at or below a predetermined level, the processing unit 50 (controlling unit 50b) considers the user’s health status to be “fatigue”.

In addition, in addition to the voice recognition result of the voice recognition unit 40, the image recognition result of the image recognition unit 20 may be further used to grasp the user's health status. Alternatively, the image recognition result of the image recognition unit 20 may be used alone to grasp the user's health status. For example, when the processing unit 50 (controlling unit 50b) detects that the user blinks frequently (or the user yawns), the processing unit 50 (controlling unit 50b) considers the user's health status to be “fatigue”.

In this way, the processing unit 50 (grasp 50b) grasps the user's health status based on the language spoken by the user or based on the image recognition result, and functions as a grasp unit.

In the information database 160, information such as weather or news of the day, knowledge such as various general knowledge, information related to the user (owner) of the robot 1 (for example, gender, age, name, occupation, personality, interest, birth date, etc.) are stored. Information such as year, month and day), and related information of the robot 1 (for example, information such as model, internal structure, etc.). For example, information such as the weather and news of the day is obtained from the outside by the robot 1 through the transmitting and receiving unit 60 (communication unit) and the processing unit 50 , and stored in the information database 160 . In addition, the information database 160 also stores the relationship between the user's health status and action patterns as the action pattern table 161 .

FIG. 4 shows an example of the action pattern table 161 stored in the information database 160 . The action pattern table 161 defines the relationship between the user's health status and the action pattern of the robot 1 .

In the example shown in FIG. 4 , the relationship between the user's health status and the action pattern of the robot 1 is expressed in the form of a table. For example, row 41 of the table indicates that the user's health state of "tiredness" corresponds to the three action modes of the robot 1 . The 3 modes of action are as follows.

1) Selection and playback of content: Select the content (software) used to exert the effects of "therapy" and "hypnosis", and play the selected content (software) through the playback device.

2) Prepare bath water: prepare bath water in order to advise the user to take a bath.

3) Selection and conditioning of food and beverage recipes: select food and beverage recipes of "appetite-enhancing" and "nutrition-rich", and prepare food and beverages according to the selected recipes.

In addition, the 42nd row of the table indicates that the user's health state of "joy" corresponds to the action pattern of the robot 1 "cheering posture".

In addition, the method of expressing the relationship between the user's health state and the action pattern of the robot 1 is not limited to the method shown in FIG. 4 . The relationship between the user's health status and the action pattern of the robot 1 can be represented by any representation method.

The example of the action mode of robot 1 includes: select the content (software) that is suitable for the user's health state, and the selected content (software) is played through the playback device; Select the beverage and food recipe that is suitable for the user's health state, and Selected recipes prepare drinks and food; prepare bath water; perform jokes to please users, etc.

The processing unit 50 (behavior pattern determination unit 50c) searches the dialogue database 140 in response to the timing signal t1 output by the grasping unit 50b, thereby using the user's health status grasped by the above search to search the information database 160 (behavior pattern table 161 ). As a result, the processing unit 50 (action pattern determining means 50c) specifies the action pattern of the robot 1 that is suitable for the user's health state. For example, when the user's health status is "fatigue", the processing unit 50 (action pattern determination unit 50c) refers to the table (action pattern table 161) shown in FIG. One of the action patterns is determined as the action pattern of the robot 1 .

Here, the processing unit 50 (action pattern determining means 50c) can specify one of the three action patterns as the action pattern of the robot 1 in various ways. For example, when the priorities of the three action patterns are assigned in advance, the action patterns of the robot 1 can be determined in order from the highest priority. The order of priority may also be changed according to time periods. For example, the priority of "preparing bath water" can be set to the highest during the period from 18:00 to 22:00, and the priority can be set at 6:00 to 8:00, 11:00 to 13:00, and 17:00 to 19:00 During the period of time, the priority of "selection and preparation of food and drink recipes" is set to the highest, and in other periods of time, the priority of "selection and playback of content" is set to the highest.

In this way, the processing unit 50 (action pattern determination means 50c) determines an appropriate action pattern based on the user's health state grasped based on the grasping means 50b, and functions as a determination means.

The processing unit 50 (execution instructing unit 50g) generates a control signal corresponding to the determined action mode in response to the timing signal t2 output by the action mode determining unit 50c, and outputs the control signal to the action control unit 120.

The operation control unit 120 drives various actuators 130 based on control signals output from the processing unit 50 (execution instruction unit 50g). Thereby, the robot 1 can be operated in a desired state.

For example, when the determined action pattern is "cheering posture", the motion control unit 120 drives the actuator (actuator) that moves the "arm" of the robot 1 up and down according to the control signal output by the processing unit 50 (execution instruction unit 50g). part of device 130). Alternatively, when the determined action mode is "selection and playback of content", the motion control unit 120 drives the "finger" ( Gripping Unit) actuator (a part of the actuator 130), so that it grips the disc and inserts the disc into the playback device. For example, assume that a plurality of discs are arranged and stored in a tray in a prescribed order.

In this way, the processing unit 50 (execution instructing means 50g) instructs the action control unit 120 to execute the action pattern determined by the action pattern determining means 50c, and functions as an execution instructing means.

Alternatively, when the determined action mode is "preparing bath water", the processing unit 50 (execution instruction unit 50g) controls the remote control unit 70 to send a remote control signal to the hot water supply control device. The hot water supply control device supplies an appropriate amount of hot water of an appropriate temperature to the bathtub according to the remote control signal (or, after supplying an appropriate amount of water to the bathtub, heats the water to an appropriate temperature). In this case, the processing unit 50 (execution instructing means 50g) instructs the remote control unit 70 to execute the action pattern determined by the action pattern determining means 50c, and functions as an execution instructing means.

Alternatively, when the determined action mode is "selection and playback of content", the processing unit 50 (execution instructing unit 50g) controls the remote control unit 70 to send a remote control signal to the playback device. The playback device selects content from the disc inserted into the playback device according to the remote control signal, and plays it. When the playback device is connected to an automatic disc changer capable of inserting multiple discs, the playback device selects content from the plurality of discs according to a remote control signal and plays them. In addition, the song selection list containing all the tracks in the multi-disc can be stored in the memory of the processing unit 50, or the playback device can read the selection list of the disc from the head of the disc, and send The receiving unit 60 stores the data in the memory in the processing unit 50 . In this case, the processing unit 50 (execution instructing means 50g) instructs the remote control unit 70 to execute the action pattern determined by the action pattern determining means 50c, and functions as an execution instructing means.

FIG. 8b shows another example of the internal structure of the processing unit 50 shown in FIG. 2 . In the example shown in FIG. 8 b , the processing unit 50 (suggestion unit 50 e ) suggests the action pattern to the user in the form of voice before instructing to execute the action pattern. For example, when the determined action pattern is "prepare bath water", the processing unit 50 (suggestion unit 50e) refers to the dialog database 140 in response to the timing signal t2 output by the action pattern determination unit 50c, and generates "You look tired. Need Are you ready for bath water?" Interrogative sentences (advice sentences) are output to the voice synthesis unit 100 . The voice synthesis unit 100 synthesizes the interrogative sentence into voice. The synthesized voice is output from the voice output unit 110 .

Next, a method for generating a suggestion sentence by the suggestion unit 50e will be described with reference to FIG. 9 . The advice unit 50e has an advice sentence synthesis unit inside. The dialog database 140 has a suggested sentence format database inside, and the suggested sentence format database stores a plurality of suggested sentence formats corresponding to a plurality of suggested expressions. The "suggestion expression" mentioned here refers to the reasons (A) and The corresponding expression of the response (B).

First, the suggestion unit (suggested sentence synthesizing unit) 50e selects from the suggested sentence format database the above-mentioned " Mastered health status" matching suggested sentence format. Next, the suggestion unit (suggestion statement synthesizing unit) 50e inserts "the health status grasped" into the position A of the suggestion statement format, and inserts the "decided action mode" into the position B of the suggestion statement format, thereby synthesizing the suggestion statement . For example, when the "mastered health state" is "fatigue" and the "decided action mode" is "prepare bath water", the suggestion unit (suggestion sentence synthesis department) 50e synthesizes "You look tired, you need to prepare bath water Is it?" Suggestions like this. The suggested words are output to the voice synthesis unit 100 . The voice synthesis unit 100 synthesizes the suggested sentence into voice. The synthesized voice is output from the voice output unit 110 .

According to this mode, the processing part 50 (suggestion unit 50e) uses the dialogue database (suggestion sentence format database) 140 and the sound synthesis part 100, the sound output part 110, before instructing to execute the action pattern determined by the action pattern determination unit 50c, This mode of action is suggested to the user by voice, and functions as a suggestion unit.

The user responds to Robot 1's suggestion by either accepting the suggestion or not. For example, the user's opinion when accepting (Yes) the suggestion can be "good", "yes", "do that", etc., and the opinion when rejecting (No) the suggestion can be "no", "don't have to." ", "No need" and other answers. Patterns of such replies are stored in the dialogue database 140 in advance.

The processing part 50 (suggestion acceptance determination unit 50f) responds to the time-series signal t5 output by the suggestion unit 50e, refers to the dialog database 140 and analyzes the voice recognition result of the voice recognition unit 40, thereby judging whether the user's answer is to accept (Yes) the suggested answer or not. Accept (No) the suggested answer.

In this manner, the processing unit 50 (suggestion acceptance determination unit 50f) utilizes the voice recognition unit 40 and the dialog database 140 to determine whether the user's reply to the suggested action pattern is a reply to accept the suggested action pattern, and to play a role in the suggestion acceptance judgment. The role of the unit.

FIG. 8c shows still another example of the internal structure of the processing unit 50 shown in FIG. 2 . Before instructing to execute the determined action mode, it may be determined whether the action mode needs to be suggested to the user. For example, the action suggestion necessity table 162 shown in FIG. The necessity determination unit 50d) can determine whether the suggestion is necessary based on the value of the flag. For example, the processing unit 50 (advice necessity determination unit 50d) suggests the action pattern to the user before instructing execution of the action pattern when the flag value assigned to the action pattern is "1"; When the value of the flag of is "0", the action mode is not suggested to the user until execution of the action mode is instructed.

For example, for the "prepare bath water" action mode, it is better to require prior advice to the user. Whether the user takes a bath depends largely on the mood at the time. If the user is not required to advise the user in advance, it may appear to be imposed on others. For another example, for the action mode of "cheering gesture", it is best not to require a suggestion to the user in advance. This is because cheering after asking users one by one for instructions in advance will give people a sense of routine.

In this manner, the processing unit 50 (advice necessity determination unit 50d) uses the information database 160 (action suggestion necessity table 162) to determine whether it is necessary to suggest the action pattern to the user before instructing execution of the decided action pattern. , to play the role of a judgment unit for the necessity of a suggestion.

In a case where a time period for executing an action pattern is determined, or in a case where the action pattern is frequently executed, it is generally desirable not to suggest the action pattern to the user one by one. Conversely, in the case of an action pattern that is not usually performed, it is preferable to suggest the action pattern to the user before instructing to execute the action pattern, thereby confirming whether the user desires to execute the action pattern.

The advice necessity determination means 50d for realizing the above-mentioned function will be described with reference to FIG. 10 . The time distribution record storage unit 90 includes a time measurement unit 91 , an accumulation unit 92 , and a time distribution database 93 . The advice necessity determination unit 50d has a comparison determination unit inside. The time measurement unit 91 receives an input from the execution instruction unit 50g, measures the time when the action pattern is executed, and outputs it to the integration unit 92 . The time distribution database 93 records and saves the number of executions of each action pattern at each time, and the accumulation unit 92, when receiving an input from the time measurement unit 91 each time, records the number of times corresponding to the time measured in the time distribution database 93 . The number of executions is increased by 1 for accumulation. In this way, the time distribution record storage unit 90 stores the history information of the action patterns executed at each time. Advice necessity determination unit (comparison determination unit) 50d holds a preset value, and when receiving an input from action pattern determination unit 50c, refers to the action at that time (or period) from time distribution record storage unit 90 The number of past executions of the pattern and compare with the above preset value. The comparison determination unit determines that it is necessary to suggest the action mode when the past execution times of the action mode is less than a preset value; This mode of action need not be suggested. This determination result is output from the advice necessity determination means 50d as the determination result of the advice necessity determination means 50d.

In this manner, the advice necessity determination unit 50d judges the necessity of advice based on the time distribution of the number of times the action pattern is executed.

FIG. 5 shows a processing procedure for the robot 1 to grasp the user's health status and instruct execution of an action pattern suited to the user's health status.

Step ST1: Master the user's health status.

For example, the processing unit 50 (grasp means 50b) extracts a keyword from the voice recognition result of the voice recognition unit 40, and searches the dialogue database 140 using the keyword. As a result, the processing unit 50 (grasp means 50b) can grasp the user's health status based on the keyword.

An example of the conversation between the user and the robot 1 is shown below. Here, U represents the language spoken by the user, and S represents the language spoken by the robot 1 .

U: I'm really tired today~.

S: It looks very tired.

In this way, when the user utters keywords such as "sleepy", "tired", and "no appetite", the processing unit 50 (grasp unit 50b) considers the user's health status to be "tired".

Step ST2: Determine an action pattern suitable for the user's health state grasped in step ST1.

For example, the processing unit 50 (action pattern determining means 50c) searches the information database 160 (action pattern table 161) using the user's health status. As a result, the processing unit 50 (action pattern determination means 50c) can determine an action pattern suitable for the user's health state. The behavior pattern is preferably preset as the already inferred desired behavior of the user.

Step ST3: Before instructing to execute the action pattern determined in step ST2, it is judged whether the action pattern needs to be suggested to the user by the advice necessity determination unit 50d.

When the determination result of step ST3 is "Yes", the process proceeds to step ST4; when the determination result of step ST3 is "No", the process proceeds to step ST6.

Step ST4: Before instructing to execute the action mode determined by the suggestion unit 50e in step ST2, suggest the action mode to the user.

An example of the conversation between the user and the robot 1 is shown below. Here, U represents the language spoken by the user, and S represents the language spoken by the robot 1 .

S: You look tired. How about playing content (software) that has therapeutic effects?

U: Please.

Step ST5: Use the suggestion acceptance determination unit 50f to determine whether the user has given an acceptance response to the action pattern suggested by the robot 1 in step ST4.

When the determination result of step ST5 is "Yes", the process proceeds to step ST6; when the determination result of step ST5 is "No", the process proceeds to step ST7.

Step ST6: Use the execution instructing unit 50g to instruct to execute the action pattern determined in step ST2.

Step ST7: Save the suggested action pattern and the fact that the user did not accept the suggestion (rejection) in the information database 160 as history information.

This history information is referred to when determining the content of the future action pattern in step ST2. Modes of action that are not accepted by the user may be assigned a lower priority.

In addition, instead of step ST7 or on the basis of step ST7, when the user accepts the suggestion in step ST5, the suggested action pattern and the user's acceptance (acceptance) of the suggestion may also be stored in the information database 160 as history information. . This history information is referred to when determining the content of the future action pattern in step ST2. Action modes already assigned to users that have accepted may be assigned higher priorities.

Preferably in this way, the order of priority assigned to the suggested action modes is changed depending on whether the user accepts the suggested action modes. Thereby, the user's preference and the like can be reflected in the determination of the action pattern. As a result, it is possible to increase the actual matching ratio between the action pattern determined by the robot 1 and the user's health state.

In addition, when the user does not accept the suggestion in step ST5, the user may also make a reverse suggestion. In this case, robot 1 receives the reverse suggestion and judges whether the reverse suggestion is likely to be executed. When it is judged that the reverse suggestion may be executed, the relationship between the health state of the user and the action pattern of the robot 1 saved by the robot 1 in the update information database 160 (for example, changing the priority order of the action pattern in the table shown in FIG. 4 , or after adding a new action mode in the table shown in FIG. 4 ), instruct to execute the counter suggestion. When it is judged that the reverse suggestion is not executable, the robot 1 notifies the user that "the reverse suggestion cannot be executed". In this way, the user's preference and the like can be reflected in the determination of the action pattern by the user's reverse suggestion. As a result, it is possible to increase the actual matching ratio between the action pattern determined by the robot 1 and the user's health state.

Furthermore, step ST3 in FIG. 5 may be omitted. In this case, for all action patterns determined according to the user's health state, the user is advised before the execution of the action pattern is instructed.

In addition, steps ST3, ST4, ST5, and ST7 in FIG. 5 may be omitted. In this case, for all the action patterns determined according to the user's health status, the execution of the action pattern can be instructed immediately without waiting for the user's reply.

As described above, according to the present embodiment, the user's health status can be grasped, and an action pattern suitable for the user's health status can be determined. This frees the user from the hassle of equipping various sensors. Furthermore, users will feel that the robot is a good friend who shares their health status.

Furthermore, a form may be adopted in which the action pattern is suggested to the user before the action pattern is instructed to be executed. In this case, since the user has the final right to accept the suggestion, the robot will not be forced to accept the suggestion, and the user has a large degree of freedom in discretion. As a result, reckless actions of the robot can be prevented, and people can experience the friendliness of the robot.

According to a questionnaire survey conducted by the General Research Institute of the Japan Efficiency Association, among the ideal robots in the minds of consumers, the most popular is "pet robots that are closer to the real thing." What everyone is looking forward to is a life-like symbiotic or entertainment robot that shares a living space with humans.

It can be understood that the robot as an example of the interactive device of the present invention is a friendly and useful robot close to life. Such a robot can help the user's life and become a good friend of the user.

In addition, the content (software) played by the playback device may include at least one of video data, audio data, and lighting control data. Audio data recorded on the recording medium (DVD, etc.) can be reproduced synchronously with video data recorded on the recording medium (DVD, etc.). Furthermore, the lighting control data recorded on the recording medium (DVD etc.) can be reproduced synchronously with the playback of audio data and/or video data recorded on the recording medium (DVD, etc.). With such synchronized playback, content (software) having a highly effective "healing" effect or "hypnotic" effect can be realized.

FIG. 6 shows an example of the configuration of a playback device 2100 that enables synchronous playback of audio data and/or video data and lighting control data. By connecting the playback device 2100 with a sound output device (such as a speaker) and an image output device (such as a TV) and a lighting device, the playback device 2100 can be linked with the music and/or video provided by the recording medium, and change the lighting mode of the lighting device ( For example, at least one of the light quantity and light color of the lighting device).

The playback device 2100 includes a controller 2220 , an interface controller (I/F controller) 2230 , and a reading unit 2120 .

The controller 2220 controls the overall operation of the playback device 2100 according to a user's operation command input from the I/F controller 2230 or a control signal supplied from the decoding unit 2140 .

The I/F controller 2230 detects a user's operation (for example, a remote control signal from the remote control unit 70 ( FIG. 2 )), and outputs an operation command (for example, a play command) corresponding to the operation to the controller 2220 .

The reading unit 2120 reads information recorded on the recording medium 2110 .

A typical recording medium 2110 is a DVD (Digital Versatile Disk: Digital Versatile Disc). However, the recording medium 2110 is not limited to DVD. The recording medium 2110 may be any type of recording medium. However, in the following description, the case where the recording medium 2110 is a DVD will be described as an example. In this case, the reading unit 2120 is, for example, optical reading.

The format of data recorded on the recording medium 2110 is a format obtained by modifying the standard format of the DVD-Video standard. That is, the format in which the lighting pack (L_PCK: Lighting Pack) is newly added to the VOBU is used. The data of L_PCK is data provided for outputting lighting control data in synchronization with audiovisual data.

In order to correspond to a wide range of applications, MPEG-2 (Moving Picture Experts Group 2) specifies two methods for multiplexing any number of encoded streams and synchronously obtaining each stream for playback. These two methods are program stream (PS: Program Stream) method and transport stream (TS: Transport Stream) method. In digital storage media such as DVD, the method of program stream (PS: Program Stream) is adopted. In the following description, the program stream (PS: Program Stream) method specified in MPEG-2 is abbreviated as "MPEG-PS method"; the transport stream (TS: Transport Stream) method specified in MPEG-2 is abbreviated as "MPEG -TS way".

NV_PCK, A_PCK, V_PCK, and SP_PCK all adopt the format following the MPEG-PS method. Therefore, L_PCK also adopts the format following the MPEG-PS method.

Furthermore, the playback device 2100 further includes a stream data generating unit 2130 and a decoding unit 2140 .

The stream data generating unit 2130 generates stream data including encoded AV data and encoded lighting control data based on the output of the reading unit 2120 . The "encoded AV data" referred to here refers to data including at least one of encoded audio data and encoded video data.

The stream data generated by the stream data generating unit 2130 has a format conforming to the MPEG-PS method. Such streaming data is obtained, for example, by receiving information recorded on DVD 2120 as an RF signal, digitizing and amplifying the RF signal, and applying EFM and demodulation processing. The structure of the stream data generation unit 2130 can be the same as the well-known structure, so detailed description is omitted here.

The decoding unit 2140 includes a decomposition unit 2150 , an AV data decoding unit 2160 , an illumination control data decoding unit 2170 , an STC generation unit 2180 , and a synchronization controller (control unit) 2190 .

The decomposer 2150 receives stream data having a format conforming to the MPEG-PS method from the stream data generation unit 2130 , and decomposes the stream data into encoded AV data and encoded lighting control data. This decomposition is realized by referring to the identification code (stream_id) in the header of the PES packet. The decomposition unit 2150 is, for example, a demultiplexer (demultiplexer).

The AV data decoding unit 2160 decodes the encoded AV data to output the AV data. The term "AV data" here refers to data including at least one of audio data and video data.

The AV data decoding part 2160 includes: a video buffer 2161, which is used to temporarily store the encoded image data output by the decomposition part 2150; a video decoder 2162, which is used to decode the encoded image data to output image data; an audio buffer 2163, used to temporarily save the encoded audio data output by the decomposition unit 2150; the audio decoder 2164, used to decode the encoded audio data to output the audio data.

The illumination control data decoding unit 2170 decodes the encoded illumination control data to output the illumination control data. The "illumination control data" as used herein refers to data for controlling a plurality of pixels included in an illumination device.

The lighting control data decoding part 2170 includes: a lighting buffer 2171, which is used to temporarily store the coded lighting control data output by the decomposition part 2150; a lighting decoder 2172, which is used to decode the coded lighting control data to output the lighting control data .

The STC generation unit 2180 generates STC (Syetem Time Clock: system clock). The STC is obtained by adjusting (that is, increasing or decreasing) the frequency of the 27 MHz reference clock based on the SCR. The STC is a time used to reproduce the reference time used in data encoding when decoding encoded data.

The synchronization controller 2190 controls the AV data decoding unit 2160 and the lighting control data decoding unit 2170 so that the timing at which the AV data decoding unit 2160 outputs AV data is synchronized with the timing at which the lighting control data decoding unit 2170 outputs lighting control data.

This synchronous playback control is achieved by, for example, controlling the video decoder 2162 to output an access unit of image data when the STC matches the PTS, and controlling the audio decoder 2164 to output audio data when the STC matches the PTS. The access unit is realized by controlling the lighting decoder 2172 to output the lighting control data when the STC and the PTS match.

The synchronization controller 2190 may also control the AV data decoding unit 2160 and the lighting control data decoding unit 2170 so that the timing of decoding the AV data by the AV data decoding unit 2160 is synchronized with the timing of decoding the lighting control data by the lighting control data decoding unit 2170 .

Such synchronous playback control is performed by, for example, controlling the video decoder 2162 to decode the access unit of video data when the STC matches the DTS, and controlling the audio decoder 2164 to decode the access unit of the audio data when the STC matches the DTS. This is realized by controlling the lighting decoder 2172 to decode the access unit of the lighting control data when the STC matches the DTS.

According to this method, video data, audio data, and lighting control data can also be controlled on the basis of the timing control of the access unit output of the video data, audio data, and lighting control data, or instead of the timing control of the access unit output of the video data, audio data, and lighting control data. Decoding timing control of access unit of data, audio data, lighting control data. This is because the output timing (order) of the access unit may differ from the decoding timing (order) of the access unit. Through such control, video data, audio data, and lighting control data can be played synchronously.

The image data output by the video decoder 2162 is output to an external device (such as a TV) through the NTSC encoder 2200 . The video decoder 2162 and the TV may be directly connected through the output terminal 2240 of the playback device 2100, or may be indirectly connected through a home LAN.

The sound data output from the audio decoder 2164 is output to an external device such as a speaker through a digital/analog converter (DAC) 2210 . The audio decoder 2164 and the speaker may be directly connected through the output terminal 2250 of the playback device 2100, or may be indirectly connected through a home LAN.

The lighting control data output from the lighting decoder 2172 is output to an external device (such as a lighting device). The lighting decoder 2172 and the lighting device may be directly connected through the output terminal 2260 of the playback device 2100, or may be indirectly connected through a home LAN.

In addition, the stream data generated by the stream data generating unit 2130 may include coded sub-video data or navigation data. For example, when the stream data includes encoded sub-video data and navigation data, the decomposition unit 2150 decomposes the stream data into encoded sub-video data and navigation data. Although not shown in FIG. 6 , the decoding unit 2140 may further include a navipack circuit, a sub-video decoder, and a closed caption data decoder. The navigation packet circuit generates a control signal by processing the navigation data, and outputs the control signal to the controller 2220 . The sub-picture decoder decodes the encoded sub-picture data, and outputs the sub-picture data to the NTSC encoder 2200 . The closed caption data decoder outputs the closed caption data to the NTSC encoder 2200 by decoding the coded closed caption data included in the coded video data. The functions of these circuits are well known and irrelevant to the subject matter of the present invention, so detailed description is omitted. In this way, the decoding unit 2140 may include well-known structures not shown in FIG. 6 .

As described above, using the playback device 2100 shown in FIG. 6, a playback device capable of simultaneously playing the lighting control data recorded in the recording medium while playing the audio data and/or image data recorded in the recording medium is provided. . By connecting an audio output device (such as a speaker), an image output device (such as a TV) and a lighting device to the playback device, the lighting mode can be changed in conjunction with the music and/or video provided by the recording medium. A lighting mode with a "healing" effect is, for example, a lighting effect in which sunlight penetrates through branches and leaves.

Industrial Applicability

As described above, with the interactive device of the present invention, the user's health status can be grasped, and an action pattern suitable for the user's health status can be determined. This frees the user from the hassle of equipping various sensors. Furthermore, the user will feel that the interactive device is a good friend who shares worries about his health. As a result, the value of the interactive device increases, and the user's sense of satisfaction and desire for the interactive device increase.

Claims (14)

1. conversational device, it has:

Grasp the unit, be used for grasping user's health status;

The decision unit is used for the action model that the above-mentioned user's that grasped according to above-mentioned grasp unit health status decision adapts with it;

Carry out indicating member, be used to refer to the action model that execution is determined by above-mentioned decision unit;

The suggestion unit is used for advising above-mentioned action model by sound to above-mentioned user before the action model that is determined by above-mentioned decision unit is carried out in indication;

Identifying unit judges to the answer of above-mentioned action model of advising whether it has accepted above-mentioned action model of advising according to above-mentioned user,

When judged result thought that above-mentioned user's answer is the answer of having accepted above-mentioned action model of advising, above-mentioned action model of advising was carried out in above-mentioned execution indicating member indication.

2. conversational device as claimed in claim 1, above-mentioned user's health status is grasped in above-mentioned grasp unit according to above-mentioned user's language performance.

3. conversational device as claimed in claim 2, above-mentioned grasp unit are grasped above-mentioned user's health status according to the said key word of above-mentioned user.

4. conversational device as claimed in claim 1, it further has the suggestion necessity and has or not identifying unit, is used for judging whether to be necessary to advise above-mentioned action model to above-mentioned user before the action model that is determined by above-mentioned decision unit is carried out in indication;

Before thinking to indicate the above-mentioned action model of execution, judged result is necessary that when above-mentioned user advises above-mentioned action model above-mentioned suggestion unit is advised above-mentioned action model by sound to above-mentioned user.

5. conversational device as claimed in claim 4, above-mentioned suggestion necessity have or not identifying unit to judge suggestion whether it is necessary or not property according to allocating the value of statistical indicant that has or not to the expression of above-mentioned action model suggestion necessity in advance.

6. conversational device as claimed in claim 4, above-mentioned suggestion necessity have or not the time of the execution number of times that identifying unit is performed according to above-mentioned action model to distribute to judge suggestion whether it is necessary or not property.

7. conversational device as claimed in claim 1, above-mentioned decision unit adapts one of a plurality of action models decision that has been assigned with priority respectively for the health status with above-mentioned user action model, whether accept this action pattern according to above-mentioned user, adjusted the priority of distributing to this action pattern.

8. conversational device as claimed in claim 1, it further has storage unit, is used for storing the action model that the health status with above-mentioned user adapts;

Above-mentioned decision unit uses the action model of storing in the said memory cells to decide above-mentioned action model.

9. conversational device as claimed in claim 1, the selection of the content that comprises in the playback equipment to be play to the action model of above-mentioned user suggestion by above-mentioned suggestion unit.

10. conversational device as claimed in claim 9, foregoing comprise voice data, image data, illumination control data; Above-mentioned playback equipment changes the light quantity of lighting device and at least one item in the light color according to the illumination control data.

11. conversational device as claimed in claim 1, above-mentioned conversational device has at least one function in agent functionality and the locomotive function.

12. conversational device as claimed in claim 1, above-mentioned user's health status are represented at least one aspect in above-mentioned user's emotion and above-mentioned user's the health status.

13. a conversational device, it has:

The sound input part, being used for the sound mapping that the user sends is voice signal;

Voice recognition portion discerns the said language of user based on the voice signal that the tut input part is exported;

Dialog database is used for logining the language that above-mentioned user can say in advance, and preserves the corresponding relation of above-mentioned language that is logged and above-mentioned user's health status;

Grasp the unit, the language of logining in language that will be obtained by the identification of tut identification part and the above-mentioned dialog database contrasts, and determines above-mentioned user's health status according to this results of comparison, grasps above-mentioned user's health status thus;

The decision unit, based on the action model table of the corresponding relation of the action model of health status of having preserved above-mentioned user and conversational device, the action model that decides the above-mentioned user's who is grasped with above-mentioned grasp unit health status to adapt;

Carry out indicating member, be used to refer to the action model that execution is determined by above-mentioned decision unit;

The suggestion unit was used for before the action model that is determined by above-mentioned decision unit is carried out in indication, based on the output result of above-mentioned grasp unit and the synthetic suggestion of the output result statement of above-mentioned decision unit, advised above-mentioned action model by sound to above-mentioned user;

Identifying unit judges to the answer of above-mentioned action model of advising whether it has accepted above-mentioned action model of advising according to above-mentioned user;

When judged result thought that above-mentioned user's answer is the answer of having accepted above-mentioned action model of advising, above-mentioned action model of advising was carried out in above-mentioned execution indicating member indication.

14. conversational device as claimed in claim 13, it further has:

Accept the unit, be used for accepting the action model that above-mentioned user oppositely advises at above-mentioned action model of advising;

Be used for judging that can above-mentioned conversational device carry out the unit of the action model of above-mentioned reverse suggestion; And

When judged result thinks that above-mentioned conversational device can be carried out the action model of above-mentioned reverse suggestion, the unit that the corresponding relation of the action model of the above-mentioned user's that preserves in the above-mentioned action model table health status and above-mentioned conversational device is upgraded.

Images